Nonlinear predictive filter for satellite attitude estimation using star sensors

The filtering process of satellite attitude/attitude-rate estimation is discussed using only star sensors. Considering the characteristics of the vector observation, a real-time predictive filter for three-axis attitude estimation is given. An improved algorithm, which simplifies the restrictive condition, is presented based on the predictive filter. The nonlinear filter accurately estimates satellite attitude by predicting uncertainty in the dynamics equation one step ahead. The simulation results demonstrated both the feasibility and robustness of the filter.